AI‑Driven University Quality Assessment across Libya Using Digital Data Sources

Abstract

this paper is prepared to build an end-to-end unsupervised analyzer using Python. This reproducible model on hand is developed based on a human‑centered, data‑driven assessment of Libyan universities that combines structural indicators as a data sample (student counts, number of faculties, teaching staff, research, and quality scores) with a real‑time sentiment signal derived from public online sources in Arabic and English. After cleaning and standardizing all features, the institution‑level sentiment is estimated and join it with structural metrics. An unsupervised clustering pipeline using (k‑means algorithm) reveals three distinct institutional profiles. Enhanced visualizations (Figs. 8–10) make the findings accessible to decision‑makers [see appendix I. Results show substantial variation in both structural performance and public perception and illustrate how perception data complements traditional indicators. The approach offers a scalable framework that can be continuously updated for monitoring higher‑education quality in Libya [See the Appendix I for the entire code]